Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Tübingen AI Center, University of Tübingen
2
0
4
6
A stealthy adversarial patch can hijack a Computer Use Agent's visual attention, forcing it to choose attacker-specified products.
LVLMs can be subtly backdoored with manipulated images, allowing attackers to inject targeted messages into multi-turn conversations and manipulate users after a specific trigger.